Kobe University Repository : Kernel

タイトルTit le

XPS and STM study of TiO2(110)-(1 x 1) surfaces immersed insimulated body fluid

著者Author(s) Sasahara, Akira / Murakami, Tatsuya / Tomitori, Masahiko

掲載誌・巻号・ページCitat ion Surface Science,668:61-67

刊行日Issue date 2018-02

資源タイプResource Type Journal Art icle / 学術雑誌論文

版区分Resource Version author

権利Rights

© 2017 Elsevier B.V. This manuscript version is made available underthe CC-BY-NC-ND 4.0 license ht tp://creat ivecommons.org/licenses/by-nc-nd/4.0/

DOI 10.1016/j.susc.2017.10.022

JaLCDOI

URL http://www.lib.kobe-u.ac.jp/handle_kernel/90004619

PDF issue: 2020-12-12

1

XPS and STM Study of TiO2(110)-(11) Surfaces

Immersed in Simulated Body Fluid

Akira Sasahara,* Tatsuya Murakami, and Masahiko Tomitori

Japan Advanced Institute of Science and Technology, Nomi, Ishikawa 923-1292, Japan

*To whom correspondence should be addressed.

E-mail: sasahara@harbor.kobe-u.ac.jp

Present address: Department of Chemistry, Faculty of Science

Kobe University

Nada-ku, Kobe 657-8501, Japan

Tel: +81-78-803-5674

Fax: +81-78-803-5674

2

Abstract

Rutile titanium dioxide (TiO2) (110)-(1×1) surfaces prepared in ultrahigh vacuum (UHV) were

removed from the UHV, immersed in Hank’s buffered salt solution (HBSS), and then re-introduced

to the UHV to be examined for an atomic-level study of the osteoconductivity of TiO2. X-ray

photoelectron spectroscopy (XPS) showed the adsorption of phosphate ions and divalent Ca ions from

the HBSS. The density of the phosphate ions increased with the immersion time for up to 1 min and

remained constant with further immersion time up to 1 week. Unlike in the case of the phosphate

ions, the density of the Ca ions decreased with the immersion time after reaching the maximum value

at 1 min immersion. Consequently, calcium phosphate was not formed on the (1×1) surface in the

1-week immersion. Scanning tunneling microscopy revealed molecule-sized spots covering the

surfaces immersed for 1 min and longer. On the basis of density analysis by XPS, the spots were

hypothesized to represent phosphate ions. Phosphate ions on the (1×1) surface are unable to

undergo arrangement needed to facilitate the epitaxial growth of the brushite.

Keywords: TiO2; calcium phosphate; STM

3

Introduction

Osteoconductivity is the ability for the materials to bond to bones. Implants bond to bones via

the hydroxyapatite layer. This process is characteristic of Ti and its alloys, which are unique

materials for replacement implants [1]. Ti-based materials are naturally coated with an oxide layer.

Hence, titanium oxide is responsible for the osteoconductivity of the Ti-based materials.

Hydroxyapatite is generally known to form via the calcium phosphate precursor phase, but the type

and formation mechanism of the calcium phosphate have not been fully explained [2]. Clarifying

the process of calcium phosphate formation would thus aid in identifying the origin of the

osteoconductivity and hence allow the enhancement of the rate and/or strength of the implant–bone

bonding.

Ad-species from a body fluid have been investigated by immersing titanium dioxide (TiO2) in

simulated body fluids, which are simulated balanced electrolyte solutions with ion concentrations

similar to those of human plasma. Hanawa et al. examined calcium phosphate layers formed on

TiO2 films covering Ti plates [3,4]. X-ray photoelectron spectroscopy (XPS) analysis revealed a

higher concentration of hydrogen phosphate ions (HPO42−) in the calcium phosphate layer at its initial

stage of growth. The authors proposed that the adsorption of the hydrogen phosphate ions and

subsequent uptake of the Ca ions by the adsorbed hydrogen phosphate ions initiate the formation of

calcium phosphate. The high ratio of hydrogen phosphate ions in the calcium phosphate layer

agrees with the results of secondary-ion mass spectroscopy reported by Chusuei et al [5]. In this

study, the authors examined the ratios of the PO3− and PO2

− ions that reflect the composition of the

calcium phosphate and concluded that dicalcium phosphate dihydrate (CaHPO4·2H2O), known as

brushite, is a predominant form of calcium phosphate on the TiO2 films. On the basis of results of

depth analysis by Auger electron spectroscopy, the authors propose Ca-mediated bridging between

O atoms of the surface and of hydrogen phosphate ions. Kokubo et al. reported that acid treatment

followed by air-annealing results in a positively charge of the Ti plates and the promoted formation

of hydroxyapatite thereon [6]. According to their model, the positively charged TiO2 surface attracts

phosphate ions (PO43−), which then lead to a negative charge of the TiO2 surface and to the attraction

of Ca and thus formation of amorphous calcium phosphate.

An atomically well-defined TiO2 surface is expected to highlight the effect of microscopic surface

structures on calcium phosphate formation. The (110) surface of rutile TiO2 exhibits such a well-

defined structure after sputter-anneal cleaning in ultrahigh vacuum (UHV) [7]. Figure 1 shows the

model of the clean (110) surface with a (1×1) structure corresponding to a simple truncation of the

4

bulk crystal [8]. Oxygen atoms that bridge two Ti atoms, known as bridging O atoms (Ob atoms),

form protruding rows along the [001] direction. Some of the Ob atoms are part of OH groups (OHb

groups), and some are missing (O vacancies) [9]. The OHb groups are formed by the dissociation

of residual H2O molecules in the UHV chamber. At the troughs between the Ob atom rows, Ti atoms

coordinated to five O atoms (Ti5c atoms) are exposed. Oxygen atoms surrounding the Ti5c atoms

are called in-plane O atoms. The OHt group is another type of OH group terminating the Ti5c atom.

It forms via the reaction of the OHb group with O adatom (Oa atom) [10] and is absent on the (1×1)

surface prepared in the UHV.

Figure 1b and 1c show empty-state images of the (1×1) surface obtained by UHV-scanning

tunneling microscopy (STM). The surface consists of flat terraces separated by monatomic steps

0.33 nm high. The bright lines along the [001] direction correspond to the Ti5c atom rows, and the

short lines bridging the adjacent Ti5c atom rows along the [11_

0] direction are either OHb groups or O

vacancies. It is impossible to classify the short lines in Figure 1c into either OHb groups or O

vacancies. The image height difference between them is ~0.03 nm even in optimized imaging

conditions [9]. In separate experiments, we determined the densities of the OHb groups and the O

vacancies prepared by the method in this study to be ~0.3 nm−2 and ~0.03 nm−2, respectively [11,12].

Hence, most of the short lines in Figure 1c represent OHb groups. The (1×1) structure has been

recently found to be retained in humid environments such as laboratory air and water [13].

In this work, we examined ad-species from the Hank’s buffered salt solution (HBSS), a commonly

used simulated body fluid, on the TiO2(110)-(1×1) surface. We found that the phosphate ions and

the Ca ions adsorbed on the (1×1) surface and that calcium phosphate did not form after 1-week

immersion. The non-formation of the calcium phosphate disagrees with the commonly known

osteoconductivity of TiO2 and suggests that the osteoconductivity is sensitive to the atomic-scale

structure of the TiO2 surface.

Experimental methods

Surface cleaning and microscope experiments were conducted using a commercial UHV-STM

system (JSPM4500S, JEOL). The system is composed of the microscope chamber and the sample

preparation chamber. The microscope chamber includes an STM stage, and the sample preparation

chamber is equipped with an Ar+ sputtering gun (EX03, Thermo) and low-energy electron diffraction

(LEED) optics (BDL600, OCI). The base pressure of the chambers was 2 × 10−8 Pa. The view

ports of the chambers were shaded unless it was necessary to view the inside.

5

Mirror-polished TiO2 wafers (7 mm × 1 mm × 0.3 mm, Shinkosha) were clamped on a Si wafer

that was used as a resistive heater. The surface was cleaned by cycles of sputtering with Ar ions and

annealing. The ion beam energy, sample current, angle of the incident beam with respect to the

surface normal, and time of the sputtering, respectively, were 2 keV, 0.5 μA, 45°, and 1 min. The

temperature and time for the annealing were 1173 K and 1 min, respectively. The temperature was

monitored at the side faces of the wafers by an infrared thermometer. Hence, the temperature of the

TiO2 surface was probably lower than the monitored temperature because of incomplete contact

between the TiO2 and Si wafers. The wafers with the (1×1) surface were yellowish white, similar

to the as-purchased wafers, but they were also very slightly bluish, indicating that the wafers were

almost stoichiometric.

The cleaned TiO2 wafers were cooled to room temperature, removed from the sample preparation

chamber, and immersed in 100 mL of HBSS (with Ca and Mg and without phenol red; Nacalai

Tesque). The perfluoroalkoxy alkane container for the HBSS was shaded and was placed at room

temperature. The pH of the HBSS was ~7.0 and increased to ~7.6 after 1 day of preservation in the

container, regardless of the presence or absence of the TiO2 wafer. For the 1 week immersion, the

HBSS was exchanged every day. After immersion, the TiO2 wafers were rinsed with Milli-Q water

for 1 min and re-introduced to the STM system. Empty-state STM images were acquired in

constant-current mode at room temperature. Electrochemically etched W wires were used as a probe.

XPS analysis was performed at room temperature using a commercial instrument (Axis Ultra DLD,

Kratos). The base pressure of the system was 6 × 10−7 Pa. The TiO2 wafers were removed from

the STM system, placed in a shaded container, and transferred in laboratory air to the XPS system.

A monochromatic Al Kα line was used as an excitation source. Charging of the TiO2 surfaces was

reduced by a low-energy electron neutralizer. The photoelectron emission angle with respect to

the perpendicular line on the surface was set to 0°. The pass energy of the analyzer and the energy

step were 160 and 1.0 eV for wide scans, respectively, and 20 and 0.1 eV for narrow scans,

respectively. The binding energy of the spectra was calibrated such that the Ti 2p3/2 peak from Ti

atoms in a bulk TiO2 crystal was 459.1 eV. With this calibration, the O 1s peak from O atoms in a

bulk TiO2 crystal (Obulk atoms) was corrected to 530.3 eV. The spectra were deconvoluted into

mixed Gaussian−Lorentzian curves (70:30) following Shirley-type background subtraction.

6

Results and discussion

A wide-scan XPS spectrum of the control, a (1×1) surface immersed in Milli-Q water for 1 day, is

shown in Figure 2 (spectrum (a)). The intense peaks at ~460 and ~530 eV are Ti 2p and O 1s peaks,

respectively. Ar and C peaks were also detected. The residue from sputtering explain the Ar atoms,

and the C atoms result from the adventitious contaminants. Spectrum (b) was obtained using the

(1×1) surface that was immersed in HBSS for 1 day. In addition to the Ti, O, Ar, and C peaks, Ca

and P peaks were observed. Elements other than P and Ca included in the HBSS, such as Cl, Mg,

S, K, and Na, were not detected, consistent with previous studies [3,4].

Figure 3 shows the narrow-scan spectra of the (1×1) surfaces that were immersed in either Milli-

Q water or HBSS. The vertical scales of the spectra were normalized with respect to the height of

the Ti 2p3/2 peak in each set of spectra. Spectra (a) were obtained from the surface that was

immersed in Milli-Q water for 1 day. The O 1s spectrum shows an intense peak resulting from the

Obulk atoms. The shoulder on the high-binding-energy side of the Obulk peak originates from low-

coordination O atoms, OH groups, OOH groups, and contaminants [13]. The Ti 2p3/2 peak lacks

the low-binding-energy shoulder originating from Ti cations with lower valence states, Tin+ (n<4)

[14], indicating that the (1×1) surface was stoichiometric. The surface O vacancies of low density

should have been healed by the dissociative adsorption of H2O molecules upon exposure to laboratory

air. The C 1s spectrum is composed of three peaks at 285.5, 286.9, and 289.0 eV, which are assigned

to C atoms involved in C–C/C–H, C–O, and O–C=O bonds (CC-C/C-H, CC-O, and CO-C=O atoms),

respectively [15]. No feature could be observed in the P 2p and Ca 2p regions.

Spectra (b) were obtained from the surface immersed in the HBSS for 1 sec. The shoulder of the

Obulk peak slightly grew at ~531.6 and ~534.0 eV. The growth of the shoulder was highlighted by

the O 1s spectrum in (a), which is represented by a gray line. The Ti 2p3/2 spectrum in (b) matches

well the spectrum in (a), thus concealing the spectrum shown in (a), which is represented by a gray

line. A single hump in the P 2p spectrum appeared at ~133.7 eV, and two humps appeared at 347.8

and 351.4 eV in the Ca 2p spectrum. A specific change in the C 1s spectrum relative to that shown

in (a) was not observed.

Spectra (c) were obtained with the surface immersed in the HBSS for 5 sec. The components of

the O 1s spectrum at 531.6 and 534.0 eV became intense. The Ti 2p3/2 spectrum remained similar

to the previous spectra and thus fit with that obtained in (a). The hump in the P 2p spectrum at 133.7

eV became a distinct peak. The spectrum was fitted by two component peaks with an intensity ratio

of 2:1 and a binding energy difference of 0.9 eV. The P atoms were in an identical chemical state

7

with a single pair of spin-orbit split peaks. The Ca 2p peaks at 347.8 and 351.4 eV became intense.

The C 1s spectrum still did not show major changes.

P atoms on the (1×1) surface are in a phosphate species form. The binding energy of the P 2p

peak at 133.7 eV was lower than that of the Ti 2p3/2 peak by 325.4 eV. This difference agrees with

that previously determined for calcium phosphates on TiO2, namely 325.2–325.4 eV [4]. According

to Connor et al [16], a major form of adsorbed phosphate species in the aqueous solution with pH of

7.0–7.6 is the phosphate ion. The components of the O 1s shoulder at 531.6 and 534.0 eV are

attributed to the phosphate ions. The binding energy differences from the Obulk peaks at 1.3 and 3.7

eV agree with the growth of the O 1s shoulder induced by the phosphate ion, which ranges from 1.0

to 4.0 eV, as reported by Zhao et al [17]. The absence of Tin+ (n < 4) shoulders in the Ti 2p3/2 peak

indicates that phosphate ion adsorption does not reduce Ti5c atoms. This independence of the Ti

2p3/2 peak shape from the adsorption of molecules has been reported upon for the adsorption of

trimethyl acetic acid [18], methanol [19], and malonic acid [20]. The Ca atoms on the (1×1) surface

are in a divalent state. The binding energy difference between the Ca 2p3/2 and Ti 2p3/2 peaks at

111.3 eV agrees with a difference of 111.2–111.3 eV for the calcium phosphates on TiO2 [4].

Extending the immersion time to 1 min and 1 hr caused no notable change in the spectra, as shown

in (d) and (e). Hence, the phosphate ions and the Ca ions were the major ad-species from the HBSS

within the 1 hr immersion. Further extension of the immersion to 1 day and 1 week induced distinct

changes in the O 1s, Ca 2p, and C 1s spectra, as shown in (f) and (g). A new peak appeared in the

O 1s spectra at ~533.1 eV. The intensity of the Ca 2p peaks decreased with the immersion time in

comparison with those in (e). The intensity of the CC-O peak increased relative to that of the CC-C/C-

H peak.

The distinct growth of the CC-O peak after the 1-day and 1-week immersions is attributed to an

increase in glucose. The binding energy difference between the CC-O and the CC-C/C-H peaks of 1.4

eV agrees with that previously reported for glucose, 1.5 eV [21]. The peak at 533.1 eV in the O 1s

spectra that appeared along with the growth of the CC-O peak is assigned to the OH groups of the

glucose.

By assuming that the P and Ca atoms were uniformly distributed on the surface, their coverages

are estimated by referring to the work of Peng et al [22]. For P atoms,

∑∞ (1)

8

where Ip/ITi is the intensity ratio of the P 2p3/2 peak with respect to the Ti 2p3/2 peak. σ and N are

the photoionization cross section and the density of the atoms, respectively. σ values of 0.789, 3.35,

and 5.22 were used for P 2p3/2, Ca 2p3/2, and Ti 2p3/2 photoelectrons [23], respectively. NTi is the

density of Ti atoms in the TiO2 layers along the [110] axis, 10.26 nm−2. d is the distance between

the adjacent TiO2 layers, 0.33 nm. λ is the inelastic mean free path of the Ti 2p photoelectrons in

bulk TiO2, 1.54 nm [24]. The screening effect of the P and Ca atoms for the Ti 2p photoelectrons is

ignored. Figure 4 shows Np and NCa calculated by equation (1) for the surfaces with different

immersion times. Np increased with the immersion time up to 1 min to a value of ~3.8 nm−2 and

remained relatively constant for 1 week (6% fluctuation). On the other hand, NCa reached a value

of 3.2 nm-2 after 60 sec of immersion and decreased with further immersion. Compared with the

surface immersed for 60 sec, the average NCa decreased by ~15% after a 1 hr immersion and by ~74%

after a 1 week immersion.

The co-adsorption of the phosphate ions and Ca ions suggests that the charges induced by their

adsorption were balanced in the HBSS. The pH of the point of zero charge of the (11) surface is

expected to be ~4.8 on the basis of a sum-frequency-generation study of the air-annealed (110) surface

by Fitts et al [25]. Therefore, the (11) surface covered by a water layer in laboratory air is

negatively charged and is rich in the OHt group. In the HBSS that has a pH of 7.0–7.6, the phosphate

ions replace the OHt groups and bond to the Ti5c atoms [26]. One phosphate ion enhances two

negative charges when replacing one OHt group.

Ti – OHt− + PO4

3− → Ti – PO43− + OH− (R1)

One negative charge increases when one phosphate ion replaces two OHt groups.

2(Ti – OHt−) + PO4

3− → Ti – (PO43−) – Ti + 2OH− (R2)

The possible adsorption configurations of the phosphate ions are bidentate chelating and bidentate

bridging [16]. Hence, the adsorbed phosphate ions in reactions 1 and 2 are in the bidentate chelating

and bidentate bridging coordinations, respectively. When the NCa/Np ratio is roughly estimated to

be 3/4 on the surfaces after immersion for 1 hr or less, the positive charges from Ca ions are

compensated for by phosphate ions with an equal proportion of the bidentate chelating and bidentate

9

bridging forms. On the other hand, the NCa distinctly decreased in immersions lasting 1 hr and

longer. The decrease in the NCa may be explained by assuming that the bidentate bridging form

becomes dominant with the extension of the immersion time.

The distribution and arrangement of the ad-species were examined by STM. Figure 5a shows the

control (1×1) surface immersed in Milli-Q water for 1 day. The surface showed the same step-

terrace structure before the removal from UHV (Figure 1b), but the terraces were covered by

molecule-sized spots. The spots are classified into three kinds of O-containing species according to

their heights. The highest spots are the hydroperoxyl (OOH) group with an “up-conformation” in

which the OOH plane is parallel to the [001] direction, the middle spots are the OHt group, and the

lowest spots are the OOH group with a “transverse conformation” in which the OOH plane is normal

to the [001] direction [10,13]. The OHt groups and the OHb groups are formed by dissociation of

H2O molecules when the wet (1×1) surface is introduced to UHV [13]. A part of the OHb groups

react with O2 to form the OOH groups. The OHt groups and the OOH groups are bonded to the Ti5c

atoms, and some of them are arranged in a (2×1) periodicity indicated by the 0.60 nm × 0.65 nm

rectangle in the inset of the figure. The density of the O-containing species was 1.8 nm−2. Non-

uniform bright species, such as the two indicated by the arrows, are adventitious contaminants.

Figure 5b shows the (1×1) surface immersed in HBSS for 1 sec. Ragged spots with diameters

smaller than those of the O-containing species appeared. A part of the ragged spots resembled

strings extended in the [11_

0] direction, as indicated by the rectangle in the inset of the figure. The

ragged spots were shorter than the OHt groups by 0.05 nm and had a height equivalent to that of the

transverse OOH groups. Spots with a clear-cut outline on the left half of the inset are the O-

containing species. Along with the emergence of the ragged spots, the density of the O-containing

species decreased to 0.52 nm−2. It was difficult to count the ragged spots because of their fuzzy

outlines. Figure 5c shows the (1×1) surface immersed in HBSS for 5 sec. The area covered with

the ragged spots increased, and the density of the O-containing species decreased to 0.46 nm−2.

When the immersion time was extended to 1 min, the area covered with the ragged spots further

increased, and the density of the O-containing species decreased to 0.29 nm−2, as shown in Figure 5d.

The inset shows the area where the Ti5c atom rows were accidentally exposed, which indicates the

monolayer thickness of the phosphate ions. A part of the ragged spots was arranged in a (2×1)

periodicity as indicated by the rectangle. While 1 hr immersion did not cause a distinct change in

surface topography (Figure 5e), bright species with a non-uniform shape increased after the

10

immersions for 1 day and 1 week, as shown in Figure 5f and 5g. The ragged spots were observed

through the breaks of the bright species.

The ragged spots are assigned to the presence of phosphate ions. The areas covered by the ragged

spots were equivalent on the surfaces immersed in HBSS for 1 min and longer, in agreement with the

relatively constant Np on these surfaces. Np is saturated when the phosphate ions occupy the Ti5c

atoms to form a monolayer. The images of the phosphate ions occasionally changed, depending on

the tip condition. Figure 5h shows the change in the image of the phosphate ions. The phosphate

ions can be observed as spots in the upper half of the image. Some of the spots were elongated in

the [11_

0] direction, but the string-like appearance is limited. The tip condition changed in the scan

line indicated by the arrows, and the phosphate ions had a string-like appearance in the lower half of

the image. The elongated shape of the phosphate ions probably reflects the spatial distribution of

the lowest unoccupied molecular orbital (LUMO) closest to the tip. Our preliminary calculation for

using the B3LYP method and the 6-311++G** basis sets showed that the LUMO of the phosphate

ion isolated in space is localized around the O atoms. When a phosphate ion bonds to two Ti5c atoms

in the bidentate configuration, two free O atoms extend laterally in the [11_

0] direction. Hence, the

LUMO around the two O atoms likely produce the spots aligned in the [11_

0] direction, although a

more precise modeling including the TiO2 substrate is necessary in order to understand the STM

image.

The strings extended in the [11_

0] direction originate from the phosphate ions arranged in phase

with the neighbors on the adjacent Ti5c atom rows. Hence, a major part of the phosphate ions are

expected to be arranged in the (2×1) periodicity with respect to the (1×1) surface. Actually, the half-

order spots along the <001> directions emerged in the LEED patterns, as indicated by the arrows in

Figure 6. The (2×1) arrangement is known to exist in the case of carboxylate molecules (R–COO−)

[27–31].

The density of the phosphate ions is 2.6 nm−2 when they occupy the Ti5c atoms to form a (2×1)

monolayer. On the other hand, the maximum Np from the XPS analysis was 3.8 nm−2 and exceeded

the density of 2.6 nm−2. This deviation stemmed from the model used to develop equation 1, in

which the attenuation of Ti 2p3/2 photoelectrons in the phosphate ion layer is ignored. Because λ

is smaller than it actually is, the denominator of the right-hand side of equation 1 provides a value

larger than that should be in actuality, which results in an overestimation of Np and NCa.

Ca ions are unlikely to exist as ragged spots. The NCa decreased to less than half between

immersions of 1 min and 1 week. This decrease disagrees with the density of the ragged spots, even

11

if we consider only the area free of the bright species in the case of the surface immersed for 1 week.

According to the density functional theory calculations by San Miguel et al., the 3-fold hollow

consisting of two Ob atoms and one in-plane O atom is the most energetically favorable site for Ca

atoms [32]. Molecular dynamics simulation by Předota et al. [33] showed that Ca ions in aqueous

solution are stabilized at hollow sites composed of Ob atoms and O atoms of the OHt groups. Ca

ions from the HBSS are probably adsorbed at the hollow sites of O atoms. The O atoms of adsorbed

phosphate ions instead of the OHt groups may constitute the hollow sites. The Ca ions are separated

from the tip more than the topmost O atoms of the phosphate ions; therefore, electron tunneling to

the Ca ions would contribute less to the formation of the image contrast.

Bicarbonate ions (HCO3−) also form a monolayer with (2×1) periodicity [34], but are inappropriate

for assignment as the ragged spots from the HBSS. The infrared adsorption spectrum of the TiO2

film shows replacement of adsorbed carbonate species with phosphate ions after immersion in the

Na2HPO4 solution [16]. The phosphate ions are more strongly coordinated to the TiO2 surface than

are the bicarbonate ions.

The non-formation of calcium phosphate on the (1×1) surface contrasts with the formation of

calcium phosphate on the (110) surface annealed in air [35]. There could be structures specific to

the air-annealed surface that act as growth sites of calcium phosphate, although they are unclear at

present. The non-formation of calcium phosphate on the (1×1) surface is hypothetically explained

by epitaxial growth of the brushite. Brushite, a candidate precursor phase of hydroxyapatite [2,5],

consists of CaHPO4 sheets and H2O bilayers that are alternately stacked along the [010] direction, as

shown in Figure 7a. Assuming that the hydrogen phosphate ions are arranged on the (1×1) surface,

the least mismatch is attained on the (010) surface of the brushite. Figure 7b shows the (010) surface

of the CaHPO4 sheet indicated by the asterisk in Figure 7a. The hydrogen phosphate ions on the

(1×1) surface need to be arranged in the c(2×2) structure, where the ions are out of phase with the

neighboring ions in the adjacent Ti5c atom rows, and the [001] direction on the (1×1) surface is the

[100] direction of the CaHPO4 sheet. The c(2×2) arrangement of the hydrogen phosphate ions is

enabled by the migration along the Ti5c atom rows. The hydrogen phosphate ions in the c(2×2)

arrangement form a parallelogram with a short side of 0.60 nm, a long side of 0.72 nm, and a larger

interior angle of 114.8°. The parallelogram is displayed in Figure 7b. The short side and the larger

interior angle of the parallelogram match the lattice constant a and the interaxial angle of brushite,

with deviations of ~3%. However, the mismatch between the long side of the parallelogram and the

12

lattice constant c of the brushite reaches as high as 15%. This large deviation is unfavorable for the

coordination of Ca ions to the hydrogen phosphate ions.

Conclusion

Phosphate ions from the HBSS produced a monolayer on the (1×1) surface. Calcium phosphate

did not form with 1-week immersion, inconsistent with the osteoconductivity of TiO2. Arrangement

of the phosphate ions suitable for the epitaxial growth of the brushite could not be attained on the

(1×1) surface.

The above results prove that the growth of calcium phosphate on TiO2 is sensitive to lattice

matching between the two surfaces. The effects of point defects such as O vacancies, OH groups,

and Ti interstitials are possible nanostructures that affect the osteoconductivity and are subjects to be

examined. Thus, the use of a well-defined single crystal surface was demonstrated to be beneficial

in gaining a microscopic view of the osteoconductivity of TiO2.

Acknowledgment

This work was supported by the Grants-in-Aid for Scientific Research from the Japanese Society

for the Promotion of Science KAKENHI (grant numbers 26600024, 26630330, 24246014, and

16K13624).

13

References

[1] D.M. Brunette, P. Tengvall, M. Textor, P. Thomsen (Eds.), Titanium in Medicine, Springer, Berlin,

2001.

[2] S.V. Dorozhkin, M. Epple, Biological and medical significance of calcium phosphates, Angew.

Chem. Int. Ed. 41 (2002) 3130–3146.

[3] T. Hanawa, M. Ota, Characterization of surface film formed on titanium in electrolyte using XPS,

Appl. Surf. Sci. 55 (1992) 269–276.

[4] T. Hanawa, M. Ota, Calcium phosphate naturally formed on titanium in electrolyte solution,

Biomaterials, 12 (1991) 767–774.

[5] C.C. Chusuei, D.W. Goodman, M.J. Van Stipdonk, D.R. Justes, K.H. Loh, E.A. Schweikert,

Solid-liquid adsorption of calcium phosphate on TiO2, Langmuir 15 (1999) 7355–7360.

[6] T. Kokubo, D.K. Pattanayak, S. Yamaguchi, H. Takadama, T. Matsushita, T. Kawai, M.

Takemoto, S. Fujibayashi, T. Nakamura, Positively charged bioactive Ti metal prepared by simple

chemical and heat treatments, J. R. Soc. Interface 7 (2010) S503–S513.

[7] C.L. Pang, R. Lindsay, G. Thornton, Chemical reactions on rutile TiO2(110), Chem. Soc. Rev. 37

(2008) 2328−2353.

[8] H. Onishi, K. Fukui, Y. Iwasawa, Atomic-scale surface structures of TiO2(110) determined by

scanning tunneling microscopy: A new surface-limited phase of titanium oxide, Bull. Chem. Soc. Jpn.

68 (1995) 2447−2458.

[9] S. Wendt, R. Schaub, J. Matthiesen, E.K. Vestergaard, E. Wahlström, M.D. Rasmussen, P.

Thostrup, L.M. Molina, E. Lægsgaard, I. Stensgaard, B. Hammer, F. Besenbacher, Oxygen vacancies

on TiO2(110) and their interaction with H2O and O2: a combined high-resolution STM and DFT study,

Surf. Sci. 598 (2005) 226–245.

[10] Y. Du, N.A. Deskins, Z. Zhang, Z. Dohnálek, M. Dupuis, I. Lyubinetsky, Imaging consecutive

steps of O2 reaction with hydroxylated TiO2(110): identification of HO2 and terminal OH

intermediates, J. Phys. Chem. C 113 (2009) 666–671.

[11] H. Tatsumi, A. Sasahara, M. Tomitori, Lateral distribution of Li atoms at the initial stage of

adsorption on TiO2(110) surface, J. Phys. Chem. C 116 (2012) 13688–13692.

[12] A. Sasahara, S. Kitamura, H. Uetsuka, H. Onishi, Oxygen-atom vacancies imaged by a

noncontact atomic force microscope operated in an atmospheric pressure of N2 gas, J. Phys. Chem.

B 108 (2004) 15735–15737.

[13] A. Sasahara, M. Tomitori, An atomic-scale study of TiO2(110) surfaces exposed to humid

14

environments, J. Phys. Chem. C 120 (2016) 21427−21435.

[14] A. Sasahara, M. Tomitori, XPS and STM study of Nb-doped TiO2(110)-(1×1) surfaces, J. Phys.

Chem. C 117 (2013) 17680−17686.

[15] E. McCafferty, J.P. Wightman, Determination of the concentration of surface hydroxyl groups

on metal oxide films by a quantitative XPS method, Surf. Interface Anal. 26 (1998) 549–564.

[16] P.A. Connor, A.J. McQuillan, Phosphate adsorption onto TiO2 from aqueous solutions: an in situ

internal reflection infrared spectroscopic study, Langmuir 15 (1999) 2916–2921.

[17] D. Zhao, C. Chen, Y. Wang, H. Ji, W. Ma, L. Zang, J. Zhao, Surface modification of TiO2 by

phosphate: effect on photocatalytic activity and mechanism implication, J. Phys. Chem. C 112 (2008)

5993–6001.

[18] J.M. White, J. Szanyi, M.A. Henderson, Thermal chemistry of trimethyl acetic acid on TiO2(110),

J. Phys. Chem. B 108 (2004) 3592–3602.

[19] P.M. Jayaweera, E.L. Quah, H. Idriss, Photoreaction of ethanol on TiO2(110) single-crystal

surface, J. Phys. Chem. C 111 (2007) 1764–1769.

[20] K.L. Syres, A.G. Thomas, D.M. Graham, B.F. Spencer, W.R. Flavell, M.J. Jackman, V.R.

Dhanak, Adsorption and stability of malonic acid on rutile TiO2 (110), studied by near edge X-ray

absorption fine structure and photoelectron spectroscopy, Surf. Sci. 626 (2014) 14–20.

[21] S. Dahle, J. Meuthen, W. Viöl, W. Maus-Friedrichs, Adsorption of silver on glucose studied with

MIES, UPS, XPS and AFM, Appl. Surf. Sci. 284 (2013) 514–522.

[22] X.D. Peng, M.A. Barteau, Characterization of oxide layers on Mg(0001) and comparison of H2O

adsorption on surface and bulk oxides, Surf. Sci. 233 (1990) 283–292.

[23] J.H. Scofield, Hartree-slater subshell photoionization cross-sections at 1254 and 1487 eV, J.

Electron Spectrosc. Relat. Phenomena 8 (1976) 129–137.

[24] G.G. Fuentes, E. Elizalde, F. Yubero, J.M. Sanz, Electron inelastic mean free path for Ti, TiC,

TiN and TiO2 as determined by quantitative reflection electron energy-loss spectroscopy, Surf.

Interface Anal. 33 (2002) 230–237.

[25] J.P. Fitts, M.L. Machesky, D.J. Wesolowski, X. Shang, J.D. Kubicki, G.W. Flynn, T.F. Heinz,

K.B. Eisenthal, Second-harmonic generation and theoretical studies of protonation at the water/α-

TiO2 (110) interface, Chem. Phys. Lett. 411 (2005) 399–403.

[26] K.E. Healy, P.D. Ducheyne, Hydration and preferential molecular adsorption on titanium in vitro,

Biomaterials 13 (1992) 553–561.

15

[27] Z.-T. Wang, J.C. Garcia, N.A. Deskins, I. Lyubinetsky, Ability of TiO2(110) surface to be fully

hydroxylated and fully reduced, Phys. Rev. B 92 (2015) 081402(R) 1−5.

[28] J.S. Prauzner-Bechcicki, S. Godlewski, A. Tekiel, P. Cyganik, J. Budzioch, M. Szymonski,

High-resolution STM studies of terephthalic acid molecules on rutile TiO2(110)-(1×1) surfaces, J.

Phys. Chem. C 113 (2009) 9309–9315.

[29] T. Qiu, M.A. Barteau, STM study of glycine on TiO2(110) single crystal surfaces, J. Coll. Int.

Sci. 303 (2006) 229–235.

[30] A. Sasahara, H. Uetsuka, T. Ishibashi, H. Onishi, The dependence of scanning tunneling

microscope topography of carboxylates on their terminal groups, J. Phys. Chem. B 107 (2003) 13925–

13928.

[31] Q. Guo, E.M. Williams, The effect of adsorbate-adsorbate interaction on the structure of

chemisorbed overlayers on TiO2(110), Surf. Sci. 433–435 (1999) 322–326.

[32] M.A. San Miguel, J. Oviedo, J.F. Sanz, Ca deposition on TiO2(110) surfaces: insights from

quantum calculations, J. Phys. Chem. C 113 (2009) 3740–3745.

[33] M. Předota, Z. Zhang, P. Fenter, D.J. Wesolowski, P.T. Cummings, Electric double layer at the

rutile (110) surface. 2. adsorption of ions from molecular dynamics and X-ray experiments, J. Phys.

Chem. B 108 (2004) 12061–12072.

[34] A. Song, E.S. Skibinski, W.J.I. DeBenedetti, A.G. Ortoll-Bloch, M.A. Hines, Nanoscale

solvation leads to spontaneous formation of a bicarbonate monolayer on rutile (110) under ambient

conditions: implications for CO2 photoreduction, J. Phys. Chem. C 120 (2016) 9326−9333.

[35] M. Murphy, M.S. Walczak, H. Hussain, M.J. Acres, C.A. Muryn, A.G. Thomas, N. Silikas, R.

Lindsay, An ex situ study of the adsorption of calcium phosphate from solution onto TiO2(110) and

Al2O3(0001), Surf. Sci. 646 (2016) 146–153.

16

Figure captions

Figure 1. (a) A ball-and-stick model of the TiO2(110)-(1×1) surface. Light-blue balls and red

balls represent Ti atoms and O atoms, respectively. The smallest gray balls bound to the O atoms

represent H atoms. The OHt group terminating the Ti5c atom is added for reference. The unit cell

indicated by the dotted line has dimensions of 0.30 nm × 0.65 nm. (b,c) STM images of the (1×1)

surface. Scan sizes are (b) 40 nm × 40 nm and (c) 5 nm × 5 nm. Sample bias voltage, +1.4 V;

tunneling current, 0.2 nA.

Figure 2. Wide-scan XPS spectra of the (11) surfaces immersed in (a) Milli-Q water and (b) HBSS.

The immersion period was 1 day for both surfaces.

Figure 3. Sets of O 1s, Ti 2p3/2, Ca 2p, C 1s, and P 2p XPS spectra of the (11) surfaces. (a) The

surface immersed in Milli-Q water for 1 day. (b–g) The surfaces immersed in HBSS for (b) 1 sec,

(c) 5 sec, (d) 1 min, (e) 1 hr, (f) 1 day, and (g) 1 week. The O 1s and Ti 2p3/2 spectra in (b–g) are

superimposed onto those in (a) represented by a gray line. The thin black lines in the P 2p spectrum

in (c) represent two fitted component peaks and an envelope curve formed by them.

Figure 4. Dependence of the densities of P (open circles) and Ca (solid circles) atoms on the (11)

surfaces on the immersion time in HBSS.

Figure 5. (a–g) STM images of the (11) surfaces with a size of 40 nm × 40 nm with close-ups of

5 nm × 5 nm. (a) The surface immersed in Milli-Q water for 1 day. (b–g) The surfaces immersed

in HBSS for (b) 1 sec, (c) 5 sec, (d) 1 min, (e) 1 hr, (f) 1 day, and (g) 1 week. The squares in the

insets in (a) and (d) indicate (2×1) unit cells. The number (i) signifies the OOH group with “up-

conformation” (ii) is the OHt group, and (iii) is OOH groups with “transverse conformation”. (h)

The surface immersed in HBSS for 1 min (5 nm × 5 nm). The scan direction is right to left and top

to bottom. The arrows indicate the scan line where the appearance of the spots accidentally changed.

Sample bias voltage, +1.6 V; tunneling current, 0.2 nA.

Figure 6. LEED patterns of the (1×1) surfaces (a) before and (b) after immersion in HBSS for 1

min. Incident electron energies for (a) and (b) are 64 eV. Dotted rectangles represent the reciprocal

unit cell of the (1×1) surface. Arrows in (b) indicate additional spots showing the (2×1) structure.

17

Figure 7. (a) A ball-and-stick model of a brushite crystal viewed from the [001] direction. Yellow,

green, red, and gray balls respectively represent Ca, P, O, and H atoms. (b) Upper part of the

CaHPO4 sheet marked with an asterisk in (a). Upper and lower panels show views from the [010]

and the [001] directions, respectively. The light blue parallelogram indicates the c(2×2) unit cell

that is formed by four phosphate ions on the (1×1) surface.

18

Figure 1. (a) A ball-and-stick model of the TiO2(110)-(1×1) surface. Light-blue balls and red

balls represent Ti atoms and O atoms, respectively. The smallest gray balls bound to the O atoms

represent H atoms. The OHt group terminating the Ti5c atom is added for reference. The unit cell

indicated by the dotted line has dimensions of 0.30 nm × 0.65 nm. (b,c) STM images of the (1×1)

surface. Scan sizes are (b) 40 nm × 40 nm and (c) 5 nm × 5 nm. Sample bias voltage, +1.4 V;

tunneling current, 0.2 nA.

(a)

~fold-coordinated li atom (Ti50 atom)

19

Figure 2. Wide-scan XPS spectra of the (11) surfaces immersed in (a) Milli-Q water and (b) HBSS.

The immersion period was 1 day for both surfaces.

1105 01s

TI

(a)

! C1s

~ 2p

Ti 3s

~ ~p "' ~. C

$ .- ,-E

(b)

Ca2s P 2s

~~ ~ 1200 1000 800 600 400 200 0

Binding energy /eV

20

Figure 3. Sets of O 1s, Ti 2p3/2, Ca 2p, C 1s, and P 2p XPS spectra of the (11) surfaces. (a) The

surface immersed in Milli-Q water for 1 day. (b–g) The surfaces immersed in HBSS for (b) 1 sec,

(c) 5 sec, (d) 1 min, (e) 1 hr, (f) 1 day, and (g) 1 week. The O 1s and Ti 2p3/2 spectra in (b–g) are

superimposed onto those in (a) represented by a gray line. The thin black lines in the P 2p spectrum

in (c) represent two fitted component peaks and an envelope curve formed by them.

O 1s

Ti2~ :ii l Ca 2p C 1s P 2p

140,000 1~.001 1500 14,000 285.5 1500 I

(a)~ ~ ~ 531.6~ ~ ~

-!:? 347.8

_A ~ (bl 53;1.o 351.4 I 133.7

~ I

U)

~ C $

a:! .s

:=:& ~ ~ ±:! ~

~ 0

~ z

~ ~ ~ ~ ~

535 530 460 350 345 290 285 135 130 Binding energy /eV

21

Figure 4. Dependence of the densities of P (open circles) and Ca (solid circles) atoms on the (11)

surfaces on the immersion time in HBSS.

4.0 § 0 ~ 0

0 • 1, 3.0

0 • • • -!:::

i20 C • ~

1.0 ! ! • •

0 10 103 103 1()4 105 106

Immersion time /s

22

Figure 5. (a–g) STM images of the (11) surfaces with a size of 40 nm × 40 nm with close-ups of

5 nm × 5 nm. (a) The surface immersed in Milli-Q water for 1 day. (b–g) The surfaces immersed

in HBSS for (b) 1 sec, (c) 5 sec, (d) 1 min, (e) 1 hr, (f) 1 day, and (g) 1 week. The squares in the

insets in (a) and (d) indicate (2×1) unit cells. The number (i) signifies the OOH group with “up-

conformation” (ii) is the OHt group, and (iii) is OOH groups with “transverse conformation”. (h)

The surface immersed in HBSS for 1 min (5 nm × 5 nm). The scan direction is right to left and top

to bottom. The arrows indicate the scan line where the appearance of the spots accidentally changed.

Sample bias voltage, +1.6 V; tunneling current, 0.2 nA.

23

Figure 6. LEED patterns of the (1×1) surfaces (a) before and (b) after immersion in HBSS for 1

min. Incident electron energies for (a) and (b) are 64 eV. Dotted rectangles represent the reciprocal

unit cell of the (1×1) surface. Arrows in (b) indicate additional spots showing the (2×1) structure.

24

Figure 7. (a) A ball-and-stick model of a brushite crystal viewed from the [001] direction. Yellow,

green, red, and gray balls respectively represent Ca, P, O, and H atoms. (b) Upper part of the

CaHPO4 sheet marked with an asterisk in (a). Upper and lower panels show views from the [010]

and the [001] directions, respectively. The light blue parallelogram indicates the c(2×2) unit cell

that is formed by four phosphate ions on the (1×1) surface.

(b)

0

· .... ~

~,,C~~~